CCL:G: What are the most efficient ways of finding an absolute energy minimum implemented in gaussian09?

[ashika torikora <ashika.torikora..gmail.com>]

Hello again,

Dear Victor,

Just to clarify, I am aware of the existence of algorithms that allow global optimization. I've personally used simulated annealing before with a molecular mechanics force field. My question was more of a practical one:

- I personally find that molecular mechanics aren't good enough to predict the global minimum. The global minimum found by a simulated annealing run using MM will most probably not correspond to a global minimum in ab initio type calculation. 

- I want practical and non-expensive ways of finding global minima in gaussian09.

Dear John,

Hyperchem's conformational search might be interesting. Is it fine enough to produce all significant conformers of a given molecule? The problem I had with Macromodel was that I wasn't able to produce all significat conformers.

＼(;ﾟ∇ﾟ)/

2015-04-10 18:57 GMT+02:00 John Keller jwkeller]-[alaska.edu <owner-chemistry%%ccl.net>:

HyperChem has a very nice automated conformation search program that uses the Clark Still MCMM algorithm. See my YouTube video which illustrates how to do this with a dimethyl oxepane.

https://www.youtube.com/watch?v=GBlojuz460Q

The limitation is that one must use MM or PM3 theory. Then as you suggest re-optimize the various minima with Gaussian DFT or ab initio.

John Keller

On Fri, Apr 10, 2015 at 2:30 AM, ashika torikora ashika.torikora_-_gmail.com <owner-chemistry[a]ccl.net> wrote:

Hello all,

I'm trying to figure out the best/easiest/least computationally expensive way of finding the absolute energy minimum of an organic molecule in gaussian09.

The way I've come to understand things is that there are 3 or 4 options:

1- Scan coordinates without optimization (using Z matrix) 

**I haven't personally tried this method yet** but I'm wondering if such a scan is really representative of what one wants to do.

2-Scan coordinates with optimization (using redundant coordinates)

I have tried this method on one coordinate but it is too computationally expensive if one wants to do 2 or 3 dihedral angles (the number of points is too large).

3-Produce a series of conformations using a conformational scan by some software and then optimize them all.

I have personally tried this using "Maestro" but have found this to be extremely unsatisfactory. In fact the geometries are rarely diverse enough and the geometry isn't explored well enough no matter the options you pick. 

4- Produce conormations manually and optimize them all.

So far this has been my method of choice. It works well but it's annoying.

So in conclusion, my questions are:

Do any of you know of a piece of software or a script that can produce/explore conformational space well enough (I suppose it shouldn't be hard to write something like this but what's the point if it already exists somewhere). 

How do you procede in order to find global minima without wasting too much time and effort?

Thank you all for reading this very long question.

＼(;ﾟ∇ﾟ)/